Growth doesn't automatically make everything more important.
As WriftAI has grown, we've added trackers, processes, and automation, and needed most of them. But some of our best decisions were the systems we chose not to build (And yes, we're hiring).
wrift.ai/blog/not-eve...
As WriftAI has grown, we've added trackers, processes, and automation, and needed most of them. But some of our best decisions were the systems we chose not to build (And yes, we're hiring).
wrift.ai/blog/not-eve...
Not Everything Needs to Scale | WriftAI
Growth creates pressure to add more systems, meetings, and processes. We've learned that some of our best decisions come from choosing not to.
wrift.ai
September 12, 2026 at 1:18 PM
Growth doesn't automatically make everything more important.
As WriftAI has grown, we've added trackers, processes, and automation, and needed most of them. But some of our best decisions were the systems we chose not to build (And yes, we're hiring).
wrift.ai/blog/not-eve...
As WriftAI has grown, we've added trackers, processes, and automation, and needed most of them. But some of our best decisions were the systems we chose not to build (And yes, we're hiring).
wrift.ai/blog/not-eve...
In rare cases a prediction gets stuck before any runtime picks it up, usually during high demand when GPUs are fully in use.
These now automatically move to a new "abandoned" status instead of hanging indefinitely.
wrift.ai/changelog/st...
These now automatically move to a new "abandoned" status instead of hanging indefinitely.
wrift.ai/changelog/st...
Stuck predictions are now marked abandoned | WriftAI
A prediction that gets stuck before a runtime picks it up is now automatically marked abandoned instead of hanging indefinitely.
wrift.ai
September 9, 2026 at 12:10 PM
In rare cases a prediction gets stuck before any runtime picks it up, usually during high demand when GPUs are fully in use.
These now automatically move to a new "abandoned" status instead of hanging indefinitely.
wrift.ai/changelog/st...
These now automatically move to a new "abandoned" status instead of hanging indefinitely.
wrift.ai/changelog/st...
Access tokens can now be created with an expiration date.
Set one when you create the token, or leave it off to keep the old behavior. Expired tokens stop working automatically, no manual revoke needed.
wrift.ai/changelog/ac...
Set one when you create the token, or leave it off to keep the old behavior. Expired tokens stop working automatically, no manual revoke needed.
wrift.ai/changelog/ac...
Access tokens can now be created with an expiration date | WriftAI
ccess tokens can now be created with an expiration date, with preset options or a custom number of days, instead of only supporting manual revocation.
wrift.ai
August 24, 2026 at 4:40 AM
Access tokens can now be created with an expiration date.
Set one when you create the token, or leave it off to keep the old behavior. Expired tokens stop working automatically, no manual revoke needed.
wrift.ai/changelog/ac...
Set one when you create the token, or leave it off to keep the old behavior. Expired tokens stop working automatically, no manual revoke needed.
wrift.ai/changelog/ac...
A model's API page now shows its full input and output reference.
Field names, types, defaults, minimums and maximums, whether something's required, and so on, laid out the same way for every model.
wrift.ai/blog/model-a...
Field names, types, defaults, minimums and maximums, whether something's required, and so on, laid out the same way for every model.
wrift.ai/blog/model-a...
Model API pages now show what a model expects and returns | WriftAI
A model's API page is now a real reference for what it supports, laid out right in the UI. No more fetching anything yourself just to find out.
wrift.ai
August 22, 2026 at 3:01 PM
A model's API page now shows its full input and output reference.
Field names, types, defaults, minimums and maximums, whether something's required, and so on, laid out the same way for every model.
wrift.ai/blog/model-a...
Field names, types, defaults, minimums and maximums, whether something's required, and so on, laid out the same way for every model.
wrift.ai/blog/model-a...
You can now cancel a running prediction straight from the dashboard.
Stuck prediction or a slow model you don't want to wait out? Hit cancel from the model's page or the prediction's detail page, no code required.
wrift.ai/changelog/ca...
Stuck prediction or a slow model you don't want to wait out? Hit cancel from the model's page or the prediction's detail page, no code required.
wrift.ai/changelog/ca...
Cancel predictions from the dashboard | WriftAI
You can now cancel a running prediction from the dashboard, not just the API, SDKs, and CLI.
wrift.ai
August 18, 2026 at 4:00 PM
You can now cancel a running prediction straight from the dashboard.
Stuck prediction or a slow model you don't want to wait out? Hit cancel from the model's page or the prediction's detail page, no code required.
wrift.ai/changelog/ca...
Stuck prediction or a slow model you don't want to wait out? Hit cancel from the model's page or the prediction's detail page, no code required.
wrift.ai/changelog/ca...
Text-to-speech isn't just text in, audio out anymore.
Modern TTS supports three modes: preset voices, voice cloning, and voice design. Each has different trade-offs in quality, control, and speed.
How to pick the right one: wrift.ai/blog/underst...
Modern TTS supports three modes: preset voices, voice cloning, and voice design. Each has different trade-offs in quality, control, and speed.
How to pick the right one: wrift.ai/blog/underst...
Understanding Modern Text to Speech | WriftAI
Modern TTS goes beyond basic speech synthesis: preset voices, voice cloning, and voice design, each with different quality, control, and speed trade-offs.
wrift.ai
August 13, 2026 at 12:38 PM
Text-to-speech isn't just text in, audio out anymore.
Modern TTS supports three modes: preset voices, voice cloning, and voice design. Each has different trade-offs in quality, control, and speed.
How to pick the right one: wrift.ai/blog/underst...
Modern TTS supports three modes: preset voices, voice cloning, and voice design. Each has different trade-offs in quality, control, and speed.
How to pick the right one: wrift.ai/blog/underst...
More dimensions doesn't mean better retrieval.
A 384-dim model can outscore a 768-dim one on the same benchmark. Pick based on your workload, not vector size.
How to pick an embedding model for RAG: wrift.ai/blog/choosin...
A 384-dim model can outscore a 768-dim one on the same benchmark. Pick based on your workload, not vector size.
How to pick an embedding model for RAG: wrift.ai/blog/choosin...
Choosing the Right Embedding Model for Your RAG Pipeline | WriftAI
How to choose an embedding model for RAG: weigh retrieval quality, dimensions, latency, and language support, not just benchmark scores.
wrift.ai
August 6, 2026 at 2:21 PM
More dimensions doesn't mean better retrieval.
A 384-dim model can outscore a 768-dim one on the same benchmark. Pick based on your workload, not vector size.
How to pick an embedding model for RAG: wrift.ai/blog/choosin...
A 384-dim model can outscore a 768-dim one on the same benchmark. Pick based on your workload, not vector size.
How to pick an embedding model for RAG: wrift.ai/blog/choosin...
Reposted by WriftAI
Great products are built by great teams.
And great teams are shaped by a handful of behaviors that show up again and again.
We wrote down a few of the ones we've come to value most at @sych.io and @wrift.ai
And great teams are shaped by a handful of behaviors that show up again and again.
We wrote down a few of the ones we've come to value most at @sych.io and @wrift.ai
The People Who Make WriftAI Better | WriftAI
We’ve learned what makes people thrive at WriftAI, from action bias and ownership to intellectual humility and teamwork.
wrift.ai
August 3, 2026 at 2:16 PM
You can now cancel a prediction while it's running.
Useful for the ones that get stuck, or just take too long, video generation, long reasoning chains, whatever. Stop it from the API, any SDK, or the CLI.
wrift.ai/blog/stop-a-...
Useful for the ones that get stuck, or just take too long, video generation, long reasoning chains, whatever. Stop it from the API, any SDK, or the CLI.
wrift.ai/blog/stop-a-...
Stop a prediction before it finishes | WriftAI
Cancel a running WriftAI prediction from the API or any SDK. Useful for stuck predictions and slow models like video generation, with no queue-time charge.
wrift.ai
July 31, 2026 at 12:30 PM
You can now cancel a prediction while it's running.
Useful for the ones that get stuck, or just take too long, video generation, long reasoning chains, whatever. Stop it from the API, any SDK, or the CLI.
wrift.ai/blog/stop-a-...
Useful for the ones that get stuck, or just take too long, video generation, long reasoning chains, whatever. Stop it from the API, any SDK, or the CLI.
wrift.ai/blog/stop-a-...
Diffusion prompts don't just describe a scene, they condition the model.
We broke down what actually moves the output: subject, environment, composition, lighting, each as its own lever instead of one vague sentence.
wrift.ai/blog/improvi...
We broke down what actually moves the output: subject, environment, composition, lighting, each as its own lever instead of one vague sentence.
wrift.ai/blog/improvi...
Improving Diffusion Outputs Through Prompt Conditioning | WriftAI
A practical guide to prompt conditioning for diffusion models: structuring prompts for predictable, controllable image outputs.
wrift.ai
July 30, 2026 at 2:05 PM
Diffusion prompts don't just describe a scene, they condition the model.
We broke down what actually moves the output: subject, environment, composition, lighting, each as its own lever instead of one vague sentence.
wrift.ai/blog/improvi...
We broke down what actually moves the output: subject, environment, composition, lighting, each as its own lever instead of one vague sentence.
wrift.ai/blog/improvi...
WriftAI now supports enterprise Single Sign-On.
SAML 2.0 and OIDC, works with Okta, Azure AD, Google, and any provider that implements either standard.
wrift.ai/blog/wriftai...
SAML 2.0 and OIDC, works with Okta, Azure AD, Google, and any provider that implements either standard.
wrift.ai/blog/wriftai...
WriftAI Now Supports Enterprise Single Sign-on | WriftAI
WriftAI now supports enterprise Single Sign-On (SSO) with SAML 2.0 and OIDC. Learn how SSO works and how to set it up for your organization's domain.
wrift.ai
July 23, 2026 at 12:53 PM
WriftAI now supports enterprise Single Sign-On.
SAML 2.0 and OIDC, works with Okta, Azure AD, Google, and any provider that implements either standard.
wrift.ai/blog/wriftai...
SAML 2.0 and OIDC, works with Okta, Azure AD, Google, and any provider that implements either standard.
wrift.ai/blog/wriftai...
Every webhook delivery now shows up in your dashboard.
If a webhook isn't arriving, you no longer have to guess whether we sent it. Filter down to just the failures and see exactly what happened.
wrift.ai/blog/webhook...
If a webhook isn't arriving, you no longer have to guess whether we sent it. Filter down to just the failures and see exactly what happened.
wrift.ai/blog/webhook...
Webhook deliveries now show up in your dashboard | WriftAI
See the status, URL, and time of every webhook we sent you, right in your dashboard. Filter down to just the failures when something is not arriving.
wrift.ai
July 15, 2026 at 12:03 PM
Every webhook delivery now shows up in your dashboard.
If a webhook isn't arriving, you no longer have to guess whether we sent it. Filter down to just the failures and see exactly what happened.
wrift.ai/blog/webhook...
If a webhook isn't arriving, you no longer have to guess whether we sent it. Filter down to just the failures and see exactly what happened.
wrift.ai/blog/webhook...
WriftAI CLI now supports OAuth 2.0 device authorization.
Log in from your terminal, approve in the browser, done. No more creating an access token and pasting it in by hand.
wrift.ai/blog/wriftai...
Log in from your terminal, approve in the browser, done. No more creating an access token and pasting it in by hand.
wrift.ai/blog/wriftai...
WriftAI CLI Gets Browser-Based Login | WriftAI
The WriftAI CLI now speaks OAuth 2.0 device authorization. Log in from your terminal, approve in your browser, done.
wrift.ai
July 14, 2026 at 12:15 PM
WriftAI CLI now supports OAuth 2.0 device authorization.
Log in from your terminal, approve in the browser, done. No more creating an access token and pasting it in by hand.
wrift.ai/blog/wriftai...
Log in from your terminal, approve in the browser, done. No more creating an access token and pasting it in by hand.
wrift.ai/blog/wriftai...
Reposted by WriftAI
The work we ship matters. So do the relationships that help us ship it.
A few reflections on team building, shared experiences, and a recent padel outing with the teams at @sych.io and @wrift.ai
wrift.ai/blog/what-we...
A few reflections on team building, shared experiences, and a recent padel outing with the teams at @sych.io and @wrift.ai
wrift.ai/blog/what-we...
What we do when we are not shipping | WriftAI
Between launches, deadlines, and product milestones, we make time for the experiences that bring our team together.
wrift.ai
July 13, 2026 at 1:09 PM
The work we ship matters. So do the relationships that help us ship it.
A few reflections on team building, shared experiences, and a recent padel outing with the teams at @sych.io and @wrift.ai
wrift.ai/blog/what-we...
A few reflections on team building, shared experiences, and a recent padel outing with the teams at @sych.io and @wrift.ai
wrift.ai/blog/what-we...
WriftAI rate limits are now documented in one place.
Previously, the only way to find out what they were was to run into one. That's no longer necessary. Everything you need is written down and linked from our docs.
wrift.ai/blog/how-to-handle-wriftai-rate-limits
Previously, the only way to find out what they were was to run into one. That's no longer necessary. Everything you need is written down and linked from our docs.
wrift.ai/blog/how-to-handle-wriftai-rate-limits
How to handle WriftAI rate limits | WriftAI
WriftAI rate limits are now documented. This post covers the limits by tier, the response headers, and how to implement correct 429 retry logic.
wrift.ai
July 1, 2026 at 9:33 PM
WriftAI rate limits are now documented in one place.
Previously, the only way to find out what they were was to run into one. That's no longer necessary. Everything you need is written down and linked from our docs.
wrift.ai/blog/how-to-handle-wriftai-rate-limits
Previously, the only way to find out what they were was to run into one. That's no longer necessary. Everything you need is written down and linked from our docs.
wrift.ai/blog/how-to-handle-wriftai-rate-limits
Starting July 1, WriftAI is switching to prepaid credits.
Auto-reload can be enabled to keep your account topped up automatically.
Existing users get a final invoice for June. Credits required from July 1.
wrift.ai/blog/we-are-switching-to-prepaid-credits
Auto-reload can be enabled to keep your account topped up automatically.
Existing users get a final invoice for June. Credits required from July 1.
wrift.ai/blog/we-are-switching-to-prepaid-credits
We are switching to prepaid credits | WriftAI
WriftAI is switching from end-of-month invoicing to prepaid credits on July 1. Here's why, how the new system works, and what to do before the switch.
wrift.ai
June 27, 2026 at 7:05 PM
Starting July 1, WriftAI is switching to prepaid credits.
Auto-reload can be enabled to keep your account topped up automatically.
Existing users get a final invoice for June. Credits required from July 1.
wrift.ai/blog/we-are-switching-to-prepaid-credits
Auto-reload can be enabled to keep your account topped up automatically.
Existing users get a final invoice for June. Credits required from July 1.
wrift.ai/blog/we-are-switching-to-prepaid-credits
Webhooks are out of beta.
You can now receive updates automatically as your model runs, instead of polling for results.
wrift.ai/changelog/we...
You can now receive updates automatically as your model runs, instead of polling for results.
wrift.ai/changelog/we...
Webhooks are now stable | WriftAI
Webhooks are out of beta. The API is stable and ready for production.
wrift.ai
June 26, 2026 at 3:40 PM
Webhooks are out of beta.
You can now receive updates automatically as your model runs, instead of polling for results.
wrift.ai/changelog/we...
You can now receive updates automatically as your model runs, instead of polling for results.
wrift.ai/changelog/we...
| We're hiring |
|______________|
\ (^_^) /
\ /
------
||
_||_
We're growing across engineering, product, sales, and marketing.
Apply at wrift.ai/careers
|______________|
\ (^_^) /
\ /
------
||
_||_
We're growing across engineering, product, sales, and marketing.
Apply at wrift.ai/careers
June 25, 2026 at 7:50 AM
| We're hiring |
|______________|
\ (^_^) /
\ /
------
||
_||_
We're growing across engineering, product, sales, and marketing.
Apply at wrift.ai/careers
|______________|
\ (^_^) /
\ /
------
||
_||_
We're growing across engineering, product, sales, and marketing.
Apply at wrift.ai/careers
We've been shipping with early partners for a few years. Until now, updates were scattered across different places.
That's fixed. Everything we ship lives at wrift.ai/changelog.
You can also subscribe to get new entries delivered to your feed reader at wrift.ai/changelog/rss.xml
That's fixed. Everything we ship lives at wrift.ai/changelog.
You can also subscribe to get new entries delivered to your feed reader at wrift.ai/changelog/rss.xml
Changelog | Features, Fixes, and Improvements | WriftAI
Your source for the latest features, improvements, and fixes for WriftAI, all in one place.
wrift.ai
June 22, 2026 at 3:13 PM
We've been shipping with early partners for a few years. Until now, updates were scattered across different places.
That's fixed. Everything we ship lives at wrift.ai/changelog.
You can also subscribe to get new entries delivered to your feed reader at wrift.ai/changelog/rss.xml
That's fixed. Everything we ship lives at wrift.ai/changelog.
You can also subscribe to get new entries delivered to your feed reader at wrift.ai/changelog/rss.xml